add docs, update colors

Author: painebenjamin

README.md

@@ -7,6 +7,10 @@
 <h2>Disk-to-GPU Tensor loading at up to 25Gbps without GDS</h2>
 </div>
 
+## Updates
+
+- **2025-11-25**: Now supports **multiple data types per checkpoint** with no regressions in speed!
+
 <div align="center">
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://github.com/fal-ai/flashpack/blob/main/media/benchmark-white.png?raw=true">
@@ -89,3 +93,95 @@ model = nn.Module(...)
 pack_to_file(model, flashpack_path)  # write state dict to file
 assign_from_file(model, flashpack_path)  # load state dict from file
 ```
+
+# CLI Commands
+
+FlashPack provides a command-line interface for converting, inspecting, and reverting flashpack files.
+
+## `flashpack convert`
+
+Convert a model to a flashpack file.
+
+```bash
+flashpack convert <path_or_repo_id> [destination_path] [options]
+```
+
+**Arguments:**
+- `path_or_repo_id` - Local path or Hugging Face repository ID
+- `destination_path` - (Optional) Output path for the flashpack file
+
+**Options:**
+| Option | Description |
+|--------|-------------|
+| `--subfolder` | Subfolder of the model (for repo_id) |
+| `--variant` | Model variant (for repo_id) |
+| `--dtype` | Target dtype for the flashpack file. When omitted, no type changes are made |
+| `--ignore-names` | Tensor names to ignore (can be specified multiple times) |
+| `--ignore-prefixes` | Tensor prefixes to ignore (can be specified multiple times) |
+| `--ignore-suffixes` | Tensor suffixes to ignore (can be specified multiple times) |
+| `--use-transformers` | Load the path as a transformers model |
+| `--use-diffusers` | Load the path as a diffusers model |
+| `-v, --verbose` | Enable verbose output |
+
+**Examples:**
+```bash
+# Convert a local model
+flashpack convert ./my_model ./my_model.flashpack
+
+# Convert from Hugging Face
+flashpack convert stabilityai/stable-diffusion-xl-base-1.0 --subfolder unet --use-diffusers
+
+# Convert with specific dtype
+flashpack convert ./my_model ./my_model.flashpack --dtype float16
+```
+
+## `flashpack revert`
+
+Revert a flashpack file back to safetensors or torch format.
+
+```bash
+flashpack revert <path> [destination_path] [options]
+```
+
+**Arguments:**
+- `path` - Path to the flashpack file
+- `destination_path` - (Optional) Output path for the reverted file
+
+**Options:**
+| Option | Description |
+|--------|-------------|
+| `-v, --verbose` | Enable verbose output |
+
+**Example:**
+```bash
+flashpack revert ./my_model.flashpack ./my_model.safetensors
+```
+
+## `flashpack metadata`
+
+Print the metadata of a flashpack file.
+
+```bash
+flashpack metadata <path> [options]
+```
+
+**Arguments:**
+- `path` - Path to the flashpack file
+
+**Options:**
+| Option | Description |
+|--------|-------------|
+| `-i, --show-index` | Show the tensor index |
+| `-j, --json` | Output metadata in JSON format |
+
+**Examples:**
+```bash
+# View basic metadata
+flashpack metadata ./my_model.flashpack
+
+# View metadata with tensor index
+flashpack metadata ./my_model.flashpack --show-index
+
+# Output as JSON
+flashpack metadata ./my_model.flashpack --json
+```

scripts/plot_benchmark.py

@@ -10,7 +10,7 @@
 # configuration
 accelerate_color = "#0f5ef3"
 flashpack_color = "#adff02"
-label_color = "#111111"
+label_color = "#eeeeee"
 model_labels = {
     "Wan-AI/Wan2.1-T2V-1.3B-Diffusers": "Wan2.1 1.3B DiT",
     "Wan-AI/Wan2.1-T2V-14B-Diffusers": "Wan2.1 14B DiT",

tests/test_speed_comparison.py

@@ -83,7 +83,7 @@ def cuda_sync():
     # Plot configuration (aligned with scripts/plot-benchmark.py)
     accelerate_color = "#0f5ef3"
     flashpack_color = "#adff02"
-    label_color = "#111111"
+    label_color = "#eeeeee"
 
     labels = [
         "pytorch",